Looking at infrared background radiation anisotropies with Spitzer II. Small scale anisotropies and their implications for new and upcoming space surveys

Abstract

Spitzer-based cosmic infrared background (CIB) fluctuations at arcminute-to-degree scales indicate the presence of new populations, whereas sub-arcminute power arises from known $z\lesssim 6$ galaxies. We reconstruct the evolution of the near-IR CIB anisotropies on sub-arcminute scales by known galaxy populations. This method is based on, and significantly advanced over, the empirical reconstruction by \cite{Helgason2012} which is combined with the halo model connecting galaxies to their host dark matter (DM) halos. The modeled CIB fluctuations from known galaxies produce the majority of the observed small-scale signal down to statistical uncertainties of $< 10\%$ and we constrain the evolution of the halo mass regime hosting such galaxies. Thus the large-scale CIB fluctuations from new populations are produced by sources with negligible small-scale power. This appears to conflict with the presented Intra-halo light (IHL) models, but is accounted for if the new sources are at high $z$. Our analysis spanning several Spitzer datasets allows us to narrow the estimated contributions of remaining known galaxies to the CIB anisotropies to be probed potentially from surveys by new and upcoming space missions such as Euclid, SPHEREx, and Roman. Of these, the Roman surveys have the best prospects for measuring the source-subtracted CIB and probing the nature of the underlying new populations at $λ<2\ μ$m, followed by Euclid's surveys, while for SPHEREx the source-subtracted CIB signal from them appears significantly overwhelmed by the CIB from remaining known galaxies.

Figures (14)

• Figure 1: Angular power spectra computed from three separate Spitzer datasets are shown. The gray points mark the new LIBRAS data Kashlinsky2025. The red and blue error bars mark the K12 data from the UDS and EGS fields respectively (K12). The green and orange data points represent the HDF-N and CDF-S data over two observational epochs (E1 and E2) respectively (KAMM4). KAMM4 points for each epoch are shifted for clarity. Errors are reported at the $1\sigma$ confidence level.
• Figure 2: Galaxy counts are shown at $m_{\rm AB} \gtrsim 18$. The shaded blue region spans the LFE and HFE models with the solid blue line being the DFE model, all of which are obtained using the best-fit Schechter parameter evolution fits from HRK12. The red diamonds indicate Spitzer SEDS galaxy counts from Ashby2013. The dashed purple and solid black lines are galaxy counts from the JWST COSMOS-Web and PEARLS surveys respectively Windhorst2023Shuntov2025cat. Errors are reported at the $1\sigma$ confidence level.
• Figure 3: Flux production rates are shown for $m_{\rm lim}=25$ and $z_{\rm max}=7$ (as chosen in HRK12). The lower and upper dashed lines represent the LFE and HFE limits of the galaxy reconstruction respectively, and the thin solid line represents the DFE, or default model. The black data points show the flux production rates computed from galaxies resolved by JWST which are $25 \leq m_{\rm AB}\lesssim 29$Kaminsky2025. There is broad agreement between the empirical model and the results from Kaminsky2025.
• Figure 4: CIB reconstructions using the HRK12 empirical model for each LIBRAS angular power spectrum (represented by the black error bars). The blue lines are of the LFE limit and the red lines the HFE limit. For each FE model, the dotted, dash-dotted, and dashed lines show contributions from the shot noise, 1-halo, and 2-halo power respectively. The shaded gray area represents the allowed range between the LFE and HFE models. The left column shows the power spectrum, the middle column shows the relative error in the model bracketing, and the right column shows the fluctuation spectrum. The average shot noise level $\bar{P}_{\rm SN}$ = mean($P_{\rm SN}^{\mathrm{LFE,HFE}}$) is shown in the left column for each power spectrum. Errors are reported at the 1$\sigma$ level.
• Figure 5: CIB reconstructions using the HRK12 empirical model for each LIBRAS angular power spectrum (represented by the black error bars). The reconstructed power spectra are shown using the same formatting as those in Fig. \ref{['fig:LIBRAS_vertical_triple_panels_li']}, with the only difference being that the shot noise levels are lower by a factor of $\sim2-3$.